Weighted Ontology for Semantic Search

This paper presents a method, SemSim, for the semantic search and retrieval of digital resources (DRs) that have been previously annotated. The annotation is performed by using a set of characterizing concepts, referred to as features, selected from a reference ontology. The proposed semantic search method requires that the features in the ontology are weighted. The weight represents the probability that a resource is annotated with the associated feature. The SemSim method operates in three stages. In the first stage, the similarity between concepts (consim) is computed by using their weights. In the second stage, the concept weights are used to derive the semantic similarity (semsim) between a user request and the DRs. In the last stage, the answer is returned in the form of a ranked list. An experiment aimed at assessing the proposed method and a comparison against a few among the most popular competing solutions is given.